From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from e28smtp03.in.ibm.com (e28smtp03.in.ibm.com [122.248.162.3]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id E41C82C009C for ; Tue, 28 Jan 2014 03:54:14 +1100 (EST) Received: from /spool/local by e28smtp03.in.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Mon, 27 Jan 2014 22:24:11 +0530 Received: from d28relay02.in.ibm.com (d28relay02.in.ibm.com [9.184.220.59]) by d28dlp03.in.ibm.com (Postfix) with ESMTP id B4BBF1258051 for ; Mon, 27 Jan 2014 22:25:52 +0530 (IST) Received: from d28av04.in.ibm.com (d28av04.in.ibm.com [9.184.220.66]) by d28relay02.in.ibm.com (8.13.8/8.13.8/NCO v10.0) with ESMTP id s0RGs6Q546924024 for ; Mon, 27 Jan 2014 22:24:07 +0530 Received: from d28av04.in.ibm.com (localhost [127.0.0.1]) by d28av04.in.ibm.com (8.14.4/8.14.4/NCO v10.0 AVout) with ESMTP id s0RGs8C2007196 for ; Mon, 27 Jan 2014 22:24:08 +0530 Received: from localhost.localdomain (dhcp-9-41-149-101.austin.ibm.com [9.41.149.101]) by d28av04.in.ibm.com (8.14.4/8.14.4/NCO v10.0 AVin) with ESMTP id s0RGs7jw007095 for ; Mon, 27 Jan 2014 22:24:08 +0530 Message-ID: <52E68F2E.4030907@linux.vnet.ibm.com> Date: Mon, 27 Jan 2014 10:54:06 -0600 From: Nathan Fontenot MIME-Version: 1.0 To: "linuxppc-dev@lists.ozlabs.org" Subject: [PATCH] powerpc/pseries Use remove_memory() to remove memory Content-Type: text/plain; charset=ISO-8859-1 List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , The memory remove code for powerpc/pseries should call remove_memory() so that we are holding the hotplug_memory lock during memory remove operations. This patch updates the memory node remove handler to call remove_memory() and adds a ppc_md.remove_memory() entry to handle pseries specific work that is called from arch_remove_memory(). During memory remove in pseries_remove_memblock() we have to stay with removing memory one section at a time. This is needed because of how memory resources are handled. During memory add for pseries (via the probe file in sysfs) we add memory one section at a time which gives us a memory resource for each section. Future patches will aim to address this so will not have to remove memory one section at a time. Signed-off-by: Nathan Fontenot --- arch/powerpc/include/asm/machdep.h | 4 + arch/powerpc/mm/mem.c | 7 + arch/powerpc/platforms/pseries/hotplug-memory.c | 85 +++++++++++------------- 3 files changed, 50 insertions(+), 46 deletions(-) Index: powerpc/arch/powerpc/platforms/pseries/hotplug-memory.c =================================================================== --- powerpc.orig/arch/powerpc/platforms/pseries/hotplug-memory.c 2014-01-27 08:47:19.000000000 -0600 +++ powerpc/arch/powerpc/platforms/pseries/hotplug-memory.c 2014-01-27 09:21:54.000000000 -0600 @@ -14,6 +14,7 @@ #include #include #include +#include #include #include @@ -75,13 +76,27 @@ } #ifdef CONFIG_MEMORY_HOTREMOVE -static int pseries_remove_memblock(unsigned long base, unsigned int memblock_size) +static int pseries_remove_memory(u64 start, u64 size) { - unsigned long start, start_pfn; - struct zone *zone; int ret; - unsigned long section; - unsigned long sections_to_remove; + + /* Remove htab bolted mappings for this section of memory */ + start = (unsigned long)__va(start); + ret = remove_section_mapping(start, start + size); + + /* Ensure all vmalloc mappings are flushed in case they also + * hit that section of memory + */ + vm_unmap_aliases(); + + return ret; +} + +static int pseries_remove_memblock(unsigned long base, unsigned int memblock_size) +{ + unsigned long block_sz, start_pfn; + int sections_per_block; + int i, nid; start_pfn = base >> PAGE_SHIFT; @@ -90,45 +105,21 @@ return 0; } - zone = page_zone(pfn_to_page(start_pfn)); - - /* - * Remove section mappings and sysfs entries for the - * section of the memory we are removing. - * - * NOTE: Ideally, this should be done in generic code like - * remove_memory(). But remove_memory() gets called by writing - * to sysfs "state" file and we can't remove sysfs entries - * while writing to it. So we have to defer it to here. - */ - sections_to_remove = (memblock_size >> PAGE_SHIFT) / PAGES_PER_SECTION; - for (section = 0; section < sections_to_remove; section++) { - unsigned long pfn = start_pfn + section * PAGES_PER_SECTION; - ret = __remove_pages(zone, pfn, PAGES_PER_SECTION); - if (ret) - return ret; + block_sz = memory_block_size_bytes(); + sections_per_block = block_sz / MIN_MEMORY_BLOCK_SIZE; + nid = memory_add_physaddr_to_nid(base); + + for (i = 0; i < sections_per_block; i++) { + remove_memory(nid, base, MIN_MEMORY_BLOCK_SIZE); + base += MIN_MEMORY_BLOCK_SIZE; } - /* - * Update memory regions for memory remove - */ + /* Update memory regions for memory remove */ memblock_remove(base, memblock_size); - - /* - * Remove htab bolted mappings for this section of memory - */ - start = (unsigned long)__va(base); - ret = remove_section_mapping(start, start + memblock_size); - - /* Ensure all vmalloc mappings are flushed in case they also - * hit that section of memory - */ - vm_unmap_aliases(); - - return ret; + return 0; } -static int pseries_remove_memory(struct device_node *np) +static int pseries_remove_mem_node(struct device_node *np) { const char *type; const unsigned int *regs; @@ -153,8 +144,8 @@ base = *(unsigned long *)regs; lmb_size = regs[3]; - ret = pseries_remove_memblock(base, lmb_size); - return ret; + pseries_remove_memblock(base, lmb_size); + return 0; } #else static inline int pseries_remove_memblock(unsigned long base, @@ -162,13 +153,13 @@ { return -EOPNOTSUPP; } -static inline int pseries_remove_memory(struct device_node *np) +static inline int pseries_remove_mem_node(struct device_node *np) { return -EOPNOTSUPP; } #endif /* CONFIG_MEMORY_HOTREMOVE */ -static int pseries_add_memory(struct device_node *np) +static int pseries_add_mem_node(struct device_node *np) { const char *type; const unsigned int *regs; @@ -254,10 +245,10 @@ switch (action) { case OF_RECONFIG_ATTACH_NODE: - err = pseries_add_memory(node); + err = pseries_add_mem_node(node); break; case OF_RECONFIG_DETACH_NODE: - err = pseries_remove_memory(node); + err = pseries_remove_mem_node(node); break; case OF_RECONFIG_UPDATE_PROPERTY: pr = (struct of_prop_reconfig *)node; @@ -277,6 +268,10 @@ if (firmware_has_feature(FW_FEATURE_LPAR)) of_reconfig_notifier_register(&pseries_mem_nb); +#ifdef CONFIG_MEMORY_HOTREMOVE + ppc_md.remove_memory = pseries_remove_memory; +#endif + return 0; } machine_device_initcall(pseries, pseries_memory_hotplug_init); Index: powerpc/arch/powerpc/include/asm/machdep.h =================================================================== --- powerpc.orig/arch/powerpc/include/asm/machdep.h 2014-01-27 08:47:18.000000000 -0600 +++ powerpc/arch/powerpc/include/asm/machdep.h 2014-01-27 08:48:31.000000000 -0600 @@ -279,6 +279,10 @@ #ifdef CONFIG_ARCH_RANDOM int (*get_random_long)(unsigned long *v); #endif + +#ifdef CONFIG_MEMORY_HOTREMOVE + int (*remove_memory)(u64, u64); +#endif }; extern void e500_idle(void); Index: powerpc/arch/powerpc/mm/mem.c =================================================================== --- powerpc.orig/arch/powerpc/mm/mem.c 2014-01-27 08:47:18.000000000 -0600 +++ powerpc/arch/powerpc/mm/mem.c 2014-01-27 08:48:31.000000000 -0600 @@ -139,9 +139,14 @@ unsigned long start_pfn = start >> PAGE_SHIFT; unsigned long nr_pages = size >> PAGE_SHIFT; struct zone *zone; + int ret; zone = page_zone(pfn_to_page(start_pfn)); - return __remove_pages(zone, start_pfn, nr_pages); + ret = __remove_pages(zone, start_pfn, nr_pages); + if (!ret && (ppc_md.remove_memory)) + ret = ppc_md.remove_memory(start, size); + + return ret; } #endif #endif /* CONFIG_MEMORY_HOTPLUG */